llvm.org GIT mirror llvm / 0392724 lib / Target / AMDGPU / AMDGPUMCInstLower.cpp
0392724

Tree @0392724 (Download .tar.gz)

AMDGPUMCInstLower.cpp @0392724raw · history · blame

  1
  2
  3
  4
  5
  6
  7
  8
  9
 10
 11
 12
 13
 14
 15
 16
 17
 18
 19
 20
 21
 22
 23
 24
 25
 26
 27
 28
 29
 30
 31
 32
 33
 34
 35
 36
 37
 38
 39
 40
 41
 42
 43
 44
 45
 46
 47
 48
 49
 50
 51
 52
 53
 54
 55
 56
 57
 58
 59
 60
 61
 62
 63
 64
 65
 66
 67
 68
 69
 70
 71
 72
 73
 74
 75
 76
 77
 78
 79
 80
 81
 82
 83
 84
 85
 86
 87
 88
 89
 90
 91
 92
 93
 94
 95
 96
 97
 98
 99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
//===- AMDGPUMCInstLower.cpp - Lower AMDGPU MachineInstr to an MCInst -----===//
//
// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
// See https://llvm.org/LICENSE.txt for license information.
// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
//
//===----------------------------------------------------------------------===//
//
/// \file
/// Code to lower AMDGPU MachineInstrs to their corresponding MCInst.
//
//===----------------------------------------------------------------------===//
//

#include "AMDGPUAsmPrinter.h"
#include "AMDGPUSubtarget.h"
#include "AMDGPUTargetMachine.h"
#include "MCTargetDesc/AMDGPUInstPrinter.h"
#include "MCTargetDesc/AMDGPUMCTargetDesc.h"
#include "R600AsmPrinter.h"
#include "SIInstrInfo.h"
#include "llvm/CodeGen/MachineBasicBlock.h"
#include "llvm/CodeGen/MachineInstr.h"
#include "llvm/IR/Constants.h"
#include "llvm/IR/Function.h"
#include "llvm/IR/GlobalVariable.h"
#include "llvm/MC/MCCodeEmitter.h"
#include "llvm/MC/MCContext.h"
#include "llvm/MC/MCExpr.h"
#include "llvm/MC/MCInst.h"
#include "llvm/MC/MCObjectStreamer.h"
#include "llvm/MC/MCStreamer.h"
#include "llvm/Support/ErrorHandling.h"
#include "llvm/Support/Format.h"
#include <algorithm>

using namespace llvm;

namespace {

class AMDGPUMCInstLower {
  MCContext &Ctx;
  const TargetSubtargetInfo &ST;
  const AsmPrinter &AP;

  const MCExpr *getLongBranchBlockExpr(const MachineBasicBlock &SrcBB,
                                       const MachineOperand &MO) const;

public:
  AMDGPUMCInstLower(MCContext &ctx, const TargetSubtargetInfo &ST,
                    const AsmPrinter &AP);

  bool lowerOperand(const MachineOperand &MO, MCOperand &MCOp) const;

  /// Lower a MachineInstr to an MCInst
  void lower(const MachineInstr *MI, MCInst &OutMI) const;

};

class R600MCInstLower : public AMDGPUMCInstLower {
public:
  R600MCInstLower(MCContext &ctx, const R600Subtarget &ST,
                  const AsmPrinter &AP);

  /// Lower a MachineInstr to an MCInst
  void lower(const MachineInstr *MI, MCInst &OutMI) const;
};


} // End anonymous namespace

#include "AMDGPUGenMCPseudoLowering.inc"

AMDGPUMCInstLower::AMDGPUMCInstLower(MCContext &ctx,
                                     const TargetSubtargetInfo &st,
                                     const AsmPrinter &ap):
  Ctx(ctx), ST(st), AP(ap) { }

static MCSymbolRefExpr::VariantKind getVariantKind(unsigned MOFlags) {
  switch (MOFlags) {
  default:
    return MCSymbolRefExpr::VK_None;
  case SIInstrInfo::MO_GOTPCREL:
    return MCSymbolRefExpr::VK_GOTPCREL;
  case SIInstrInfo::MO_GOTPCREL32_LO:
    return MCSymbolRefExpr::VK_AMDGPU_GOTPCREL32_LO;
  case SIInstrInfo::MO_GOTPCREL32_HI:
    return MCSymbolRefExpr::VK_AMDGPU_GOTPCREL32_HI;
  case SIInstrInfo::MO_REL32_LO:
    return MCSymbolRefExpr::VK_AMDGPU_REL32_LO;
  case SIInstrInfo::MO_REL32_HI:
    return MCSymbolRefExpr::VK_AMDGPU_REL32_HI;
  case SIInstrInfo::MO_ABS32_LO:
    return MCSymbolRefExpr::VK_AMDGPU_ABS32_LO;
  case SIInstrInfo::MO_ABS32_HI:
    return MCSymbolRefExpr::VK_AMDGPU_ABS32_HI;
  }
}

const MCExpr *AMDGPUMCInstLower::getLongBranchBlockExpr(
  const MachineBasicBlock &SrcBB,
  const MachineOperand &MO) const {
  const MCExpr *DestBBSym
    = MCSymbolRefExpr::create(MO.getMBB()->getSymbol(), Ctx);
  const MCExpr *SrcBBSym = MCSymbolRefExpr::create(SrcBB.getSymbol(), Ctx);

  // FIXME: The first half of this assert should be removed. This should
  // probably be PC relative instead of using the source block symbol, and
  // therefore the indirect branch expansion should use a bundle.
  assert(
      skipDebugInstructionsForward(SrcBB.begin(), SrcBB.end())->getOpcode() ==
          AMDGPU::S_GETPC_B64 &&
      ST.getInstrInfo()->get(AMDGPU::S_GETPC_B64).Size == 4);

  // s_getpc_b64 returns the address of next instruction.
  const MCConstantExpr *One = MCConstantExpr::create(4, Ctx);
  SrcBBSym = MCBinaryExpr::createAdd(SrcBBSym, One, Ctx);

  if (MO.getTargetFlags() == SIInstrInfo::MO_LONG_BRANCH_FORWARD)
    return MCBinaryExpr::createSub(DestBBSym, SrcBBSym, Ctx);

  assert(MO.getTargetFlags() == SIInstrInfo::MO_LONG_BRANCH_BACKWARD);
  return MCBinaryExpr::createSub(SrcBBSym, DestBBSym, Ctx);
}

bool AMDGPUMCInstLower::lowerOperand(const MachineOperand &MO,
                                     MCOperand &MCOp) const {
  switch (MO.getType()) {
  default:
    llvm_unreachable("unknown operand type");
  case MachineOperand::MO_Immediate:
    MCOp = MCOperand::createImm(MO.getImm());
    return true;
  case MachineOperand::MO_Register:
    MCOp = MCOperand::createReg(AMDGPU::getMCReg(MO.getReg(), ST));
    return true;
  case MachineOperand::MO_MachineBasicBlock: {
    if (MO.getTargetFlags() != 0) {
      MCOp = MCOperand::createExpr(
        getLongBranchBlockExpr(*MO.getParent()->getParent(), MO));
    } else {
      MCOp = MCOperand::createExpr(
        MCSymbolRefExpr::create(MO.getMBB()->getSymbol(), Ctx));
    }

    return true;
  }
  case MachineOperand::MO_GlobalAddress: {
    const GlobalValue *GV = MO.getGlobal();
    SmallString<128> SymbolName;
    AP.getNameWithPrefix(SymbolName, GV);
    MCSymbol *Sym = Ctx.getOrCreateSymbol(SymbolName);
    const MCExpr *Expr =
      MCSymbolRefExpr::create(Sym, getVariantKind(MO.getTargetFlags()),Ctx);
    int64_t Offset = MO.getOffset();
    if (Offset != 0) {
      Expr = MCBinaryExpr::createAdd(Expr,
                                     MCConstantExpr::create(Offset, Ctx), Ctx);
    }
    MCOp = MCOperand::createExpr(Expr);
    return true;
  }
  case MachineOperand::MO_ExternalSymbol: {
    MCSymbol *Sym = Ctx.getOrCreateSymbol(StringRef(MO.getSymbolName()));
    Sym->setExternal(true);
    const MCSymbolRefExpr *Expr = MCSymbolRefExpr::create(Sym, Ctx);
    MCOp = MCOperand::createExpr(Expr);
    return true;
  }
  case MachineOperand::MO_RegisterMask:
    // Regmasks are like implicit defs.
    return false;
  }
}

void AMDGPUMCInstLower::lower(const MachineInstr *MI, MCInst &OutMI) const {
  unsigned Opcode = MI->getOpcode();
  const auto *TII = static_cast<const SIInstrInfo*>(ST.getInstrInfo());

  // FIXME: Should be able to handle this with emitPseudoExpansionLowering. We
  // need to select it to the subtarget specific version, and there's no way to
  // do that with a single pseudo source operation.
  if (Opcode == AMDGPU::S_SETPC_B64_return)
    Opcode = AMDGPU::S_SETPC_B64;
  else if (Opcode == AMDGPU::SI_CALL) {
    // SI_CALL is just S_SWAPPC_B64 with an additional operand to track the
    // called function (which we need to remove here).
    OutMI.setOpcode(TII->pseudoToMCOpcode(AMDGPU::S_SWAPPC_B64));
    MCOperand Dest, Src;
    lowerOperand(MI->getOperand(0), Dest);
    lowerOperand(MI->getOperand(1), Src);
    OutMI.addOperand(Dest);
    OutMI.addOperand(Src);
    return;
  } else if (Opcode == AMDGPU::SI_TCRETURN) {
    // TODO: How to use branch immediate and avoid register+add?
    Opcode = AMDGPU::S_SETPC_B64;
  }

  int MCOpcode = TII->pseudoToMCOpcode(Opcode);
  if (MCOpcode == -1) {
    LLVMContext &C = MI->getParent()->getParent()->getFunction().getContext();
    C.emitError("AMDGPUMCInstLower::lower - Pseudo instruction doesn't have "
                "a target-specific version: " + Twine(MI->getOpcode()));
  }

  OutMI.setOpcode(MCOpcode);

  for (const MachineOperand &MO : MI->explicit_operands()) {
    MCOperand MCOp;
    lowerOperand(MO, MCOp);
    OutMI.addOperand(MCOp);
  }
}

bool AMDGPUAsmPrinter::lowerOperand(const MachineOperand &MO,
                                    MCOperand &MCOp) const {
  const GCNSubtarget &STI = MF->getSubtarget<GCNSubtarget>();
  AMDGPUMCInstLower MCInstLowering(OutContext, STI, *this);
  return MCInstLowering.lowerOperand(MO, MCOp);
}

static const MCExpr *lowerAddrSpaceCast(const TargetMachine &TM,
                                        const Constant *CV,
                                        MCContext &OutContext) {
  // TargetMachine does not support llvm-style cast. Use C++-style cast.
  // This is safe since TM is always of type AMDGPUTargetMachine or its
  // derived class.
  auto &AT = static_cast<const AMDGPUTargetMachine&>(TM);
  auto *CE = dyn_cast<ConstantExpr>(CV);

  // Lower null pointers in private and local address space.
  // Clang generates addrspacecast for null pointers in private and local
  // address space, which needs to be lowered.
  if (CE && CE->getOpcode() == Instruction::AddrSpaceCast) {
    auto Op = CE->getOperand(0);
    auto SrcAddr = Op->getType()->getPointerAddressSpace();
    if (Op->isNullValue() && AT.getNullPointerValue(SrcAddr) == 0) {
      auto DstAddr = CE->getType()->getPointerAddressSpace();
      return MCConstantExpr::create(AT.getNullPointerValue(DstAddr),
        OutContext);
    }
  }
  return nullptr;
}

const MCExpr *AMDGPUAsmPrinter::lowerConstant(const Constant *CV) {
  if (const MCExpr *E = lowerAddrSpaceCast(TM, CV, OutContext))
    return E;
  return AsmPrinter::lowerConstant(CV);
}

void AMDGPUAsmPrinter::EmitInstruction(const MachineInstr *MI) {
  if (emitPseudoExpansionLowering(*OutStreamer, MI))
    return;

  const GCNSubtarget &STI = MF->getSubtarget<GCNSubtarget>();
  AMDGPUMCInstLower MCInstLowering(OutContext, STI, *this);

  StringRef Err;
  if (!STI.getInstrInfo()->verifyInstruction(*MI, Err)) {
    LLVMContext &C = MI->getParent()->getParent()->getFunction().getContext();
    C.emitError("Illegal instruction detected: " + Err);
    MI->print(errs());
  }

  if (MI->isBundle()) {
    const MachineBasicBlock *MBB = MI->getParent();
    MachineBasicBlock::const_instr_iterator I = ++MI->getIterator();
    while (I != MBB->instr_end() && I->isInsideBundle()) {
      EmitInstruction(&*I);
      ++I;
    }
  } else {
    // We don't want SI_MASK_BRANCH/SI_RETURN_TO_EPILOG encoded. They are
    // placeholder terminator instructions and should only be printed as
    // comments.
    if (MI->getOpcode() == AMDGPU::SI_MASK_BRANCH) {
      if (isVerbose()) {
        SmallVector<char, 16> BBStr;
        raw_svector_ostream Str(BBStr);

        const MachineBasicBlock *MBB = MI->getOperand(0).getMBB();
        const MCSymbolRefExpr *Expr
          = MCSymbolRefExpr::create(MBB->getSymbol(), OutContext);
        Expr->print(Str, MAI);
        OutStreamer->emitRawComment(Twine(" mask branch ") + BBStr);
      }

      return;
    }

    if (MI->getOpcode() == AMDGPU::SI_RETURN_TO_EPILOG) {
      if (isVerbose())
        OutStreamer->emitRawComment(" return to shader part epilog");
      return;
    }

    if (MI->getOpcode() == AMDGPU::WAVE_BARRIER) {
      if (isVerbose())
        OutStreamer->emitRawComment(" wave barrier");
      return;
    }

    if (MI->getOpcode() == AMDGPU::SI_MASKED_UNREACHABLE) {
      if (isVerbose())
        OutStreamer->emitRawComment(" divergent unreachable");
      return;
    }

    MCInst TmpInst;
    MCInstLowering.lower(MI, TmpInst);
    EmitToStreamer(*OutStreamer, TmpInst);

#ifdef EXPENSIVE_CHECKS
    // Sanity-check getInstSizeInBytes on explicitly specified CPUs (it cannot
    // work correctly for the generic CPU).
    //
    // The isPseudo check really shouldn't be here, but unfortunately there are
    // some negative lit tests that depend on being able to continue through
    // here even when pseudo instructions haven't been lowered.
    if (!MI->isPseudo() && STI.isCPUStringValid(STI.getCPU())) {
      SmallVector<MCFixup, 4> Fixups;
      SmallVector<char, 16> CodeBytes;
      raw_svector_ostream CodeStream(CodeBytes);

      std::unique_ptr<MCCodeEmitter> InstEmitter(createSIMCCodeEmitter(
          *STI.getInstrInfo(), *OutContext.getRegisterInfo(), OutContext));
      InstEmitter->encodeInstruction(TmpInst, CodeStream, Fixups, STI);

      assert(CodeBytes.size() == STI.getInstrInfo()->getInstSizeInBytes(*MI));
    }
#endif

    if (DumpCodeInstEmitter) {
      // Disassemble instruction/operands to text
      DisasmLines.resize(DisasmLines.size() + 1);
      std::string &DisasmLine = DisasmLines.back();
      raw_string_ostream DisasmStream(DisasmLine);

      AMDGPUInstPrinter InstPrinter(*TM.getMCAsmInfo(), *STI.getInstrInfo(),
                                    *STI.getRegisterInfo());
      InstPrinter.printInst(&TmpInst, DisasmStream, StringRef(), STI);

      // Disassemble instruction/operands to hex representation.
      SmallVector<MCFixup, 4> Fixups;
      SmallVector<char, 16> CodeBytes;
      raw_svector_ostream CodeStream(CodeBytes);

      DumpCodeInstEmitter->encodeInstruction(
          TmpInst, CodeStream, Fixups, MF->getSubtarget<MCSubtargetInfo>());
      HexLines.resize(HexLines.size() + 1);
      std::string &HexLine = HexLines.back();
      raw_string_ostream HexStream(HexLine);

      for (size_t i = 0; i < CodeBytes.size(); i += 4) {
        unsigned int CodeDWord = *(unsigned int *)&CodeBytes[i];
        HexStream << format("%s%08X", (i > 0 ? " " : ""), CodeDWord);
      }

      DisasmStream.flush();
      DisasmLineMaxLen = std::max(DisasmLineMaxLen, DisasmLine.size());
    }
  }
}

R600MCInstLower::R600MCInstLower(MCContext &Ctx, const R600Subtarget &ST,
                                 const AsmPrinter &AP) :
        AMDGPUMCInstLower(Ctx, ST, AP) { }

void R600MCInstLower::lower(const MachineInstr *MI, MCInst &OutMI) const {
  OutMI.setOpcode(MI->getOpcode());
  for (const MachineOperand &MO : MI->explicit_operands()) {
    MCOperand MCOp;
    lowerOperand(MO, MCOp);
    OutMI.addOperand(MCOp);
  }
}

void R600AsmPrinter::EmitInstruction(const MachineInstr *MI) {
  const R600Subtarget &STI = MF->getSubtarget<R600Subtarget>();
  R600MCInstLower MCInstLowering(OutContext, STI, *this);

  StringRef Err;
  if (!STI.getInstrInfo()->verifyInstruction(*MI, Err)) {
    LLVMContext &C = MI->getParent()->getParent()->getFunction().getContext();
    C.emitError("Illegal instruction detected: " + Err);
    MI->print(errs());
  }

  if (MI->isBundle()) {
    const MachineBasicBlock *MBB = MI->getParent();
    MachineBasicBlock::const_instr_iterator I = ++MI->getIterator();
    while (I != MBB->instr_end() && I->isInsideBundle()) {
      EmitInstruction(&*I);
      ++I;
    }
  } else {
    MCInst TmpInst;
    MCInstLowering.lower(MI, TmpInst);
    EmitToStreamer(*OutStreamer, TmpInst);
 }
}

const MCExpr *R600AsmPrinter::lowerConstant(const Constant *CV) {
  if (const MCExpr *E = lowerAddrSpaceCast(TM, CV, OutContext))
    return E;
  return AsmPrinter::lowerConstant(CV);
}